Before You Buy Business Software, Test How You Leave It

Before you sign the software contract, ask the vendor to show you what happens when you leave. Bring a customer record, an order and an attachment. Then ask someone to make that small piece of your business work somewhere else. You will learn something a polished sales demonstration rarely tells you: how much of your operation you can actually take with you.

The subscription price is easy to compare. The work of untangling your records later is much harder to see. A system can be affordable to adopt and expensive to replace because the useful context lives in its relationships, permissions and workflows, not just in the cells you can download.

Our view: for software that holds business-critical records, a demonstrated exit path should be a buying requirement. If that path is incomplete, put the reconstruction work and its cost into the decision now. Do not discover it when a price increase, service failure or change in direction has already forced your hand.

Your data can leave. Can your business still use it?

Imagine a small distributor moving to a new customer-management system. This is a hypothetical example. Every customer and order appears in the export. The totals match. Someone signs off.

Then an employee opens an order and needs the approved quote. The attachment is in a folder, but its connection to the order has gone. Two contacts share the same name. A discount appears in a column, but the approval history is missing. The team has the files and still has to rebuild the meaning.

That is the expensive part of an exit: deciding which records belong together, which history can be trusted and which work must be repeated by hand. Counting rows will not answer those questions.

The documentation already gives you reasons to look

Microsoft’s Power Apps Excel export guidance, updated October 30, 2026, says lookup fields appear as text in Excel. It also explains that calculated, roll-up and secured columns cannot be imported back through that route, even though their cells can be edited in the workbook. An editable spreadsheet can therefore give a misleading impression of how much can be reconstructed through the same route.

Google’s filtered Workspace export guidance presents a different boundary: it excludes data not owned by individual users, including shared drives. Google directs administrators to a full organizational export for shared-drive content, and available features depend on the edition. Choosing the wrong export scope can leave out material the business expected to take.

Neither example establishes a defect or a reason to reject the platform. Both show why “we support export” is too broad an answer for a purchasing decision. You need the specific route, what it includes and what the destination can do with it.

Make the vendor demonstrate one complete business record

Start small enough to inspect properly. Use synthetic data: a customer, an order, line items and a supporting attachment. Add two contacts with the same name so a successful test depends on identifiers rather than a lucky text match. Choose the destination before running the test; an export inspected in isolation cannot establish that migration will work.

A usable exit has three gates

1 · Extract

The intended records and files are present.

2 · Reconnect

IDs, relationships and attachments still resolve.

3 · Operate

A user can complete the required workflow.

The three gates keep the demonstration honest. First, retrieve the intended records and files. Then reconnect them. Finally, have an ordinary authorized user complete a real task in the destination. Opening the export on an administrator’s laptop only gets you part of the way.

Six answers worth getting before you commit

1. What is missing? Require a written list of included and excluded objects, files and date ranges. Count each record type separately.
2. What joins the records? Show the stable IDs and the mapping for parent-child links and attachments. Duplicate names should not create ambiguity.
3. What happens to history? Decide which timestamps, ownership fields and approvals must migrate. Where a protected read-only archive is acceptable, explain how staff will find and use it.
4. Who can see what? Test the destination with an ordinary user account. Restoring the records while exposing them to the wrong people is a failed migration.
5. What still needs rebuilding? Complete the chosen workflow and record every manual correction, missing feature and dependency on the old supplier.
6. What will leaving cost? Include data transformation, supplier help, overlap subscriptions and staff time. Mark estimates as estimates instead of hiding uncertainty behind a single price.

Athelguard conclusion: price the dependence before you buy

Some dependence on a supplier is a sensible trade. Useful software saves work precisely because it handles things you do not want to build yourself. The question is whether the value justifies the cost and difficulty of changing course.

We would accept an old notes archive remaining read-only if the business can retrieve what it needs. We would not call an exit ready if staff cannot reliably connect active orders to their supporting records. Those are different consequences and deserve different decisions.

Buy with a clear view of what you are depending on. If the supplier cannot demonstrate the exit you need, treat that as unresolved implementation work with an owner, a cost and a decision attached to it. A successful download should not close the discussion.

Athelguard’s Independent Technical Assessments help businesses examine technology claims, interpret the evidence and identify the checks that matter to a decision. Tell us which software you are considering and which workflow you need to preserve.

Sources checked November 8, 2026. The scenario and proposed test are Athelguard analysis, not results from hands-on testing or a completed assessment of the named products. Featured image: original AI-generated conceptual artwork, not a photograph of Microsoft or Google products.